Story perspectives

OpenAI Unveils GPT-RealTime: Emotionally Adaptive Voice Model

8/29/2025

Summary

Story summary

OpenAI launched the GPT-RealTime voice model on August 29, 2025, featuring emotional adaptability and multilingual support.
The model demonstrates an 82.8% accuracy rate in understanding complex instructions and non-verbal cues.
Developers can integrate image inputs and use SIP for telephony, priced at $32 per million audio input tokens.
OpenAI collaborates with companies like T-Mobile to enhance customer interactions through real-world applications.

History

Historical Context

Google Assistant's launch in 2016 marked a pivotal moment in voice recognition technology, significantly boosting user engagement by 90% for businesses using voice interfaces. This advancement parallels OpenAI's GPT-RealTime model, which achieved 82.8% accuracy and multilingual support, highlighting the ongoing evolution of AI in enhancing customer interactions and shaping future communication technologies.

1 of 2

Story summary

OpenAI launched the GPT-RealTime voice model on August 29, 2025, featuring emotional adaptability and multilingual support.
The model demonstrates an 82.8% accuracy rate in understanding complex instructions and non-verbal cues.
Developers can integrate image inputs and use SIP for telephony, priced at $32 per million audio input tokens.
OpenAI collaborates with companies like T-Mobile to enhance customer interactions through real-world applications.

1 / 2